Learning Forum

[Joerg]

Hello, I am looking for tips and advice for the best way to set up my simulation.

Background: I want to investigate the effect of far field soil liquefaction on soil-structure-interaction (SSI) results of a deeply embedded cylindrical structure under earthquake loading. Goal is to to conduct a parametric study, representing the structure by linear elastic elements, surrounded by non-liquefiable backfill soil (modelled nonlinear, red in screenshot attached), and liquefiable soil (modelled nonlinear, green in screenshot attached) in the far field. I would then change the extent and properties of the non-liquefiable soil until the impacts on the SSI results are negligible. Since I am only interested in the relative difference in results of the different scenarios, I was thinking a 2D analysis would be appropriate.

At this time my two main questions / concerns are:

1) To model soil liquefaction I want to follow the approach per paper by Richard Sturt 'Modelling liquefaction of soils with LS-DYNA using a SANISAND-based material model'. Now per my research so far the SANISAND material model is meant for solid elements only. So is the MAT_HYSTERETIC, which I was planning on using for the non-liqueqfiable soil. Could I use these material models in a 2D analysis ? Apparently they do work for plane strain and axisymmetric shell formulations (and thick shell formulations for MAT_HYSTERETIC)...

2) When solving a very simple 2D analysis in LS Dyna, I received the following warning message: "A 2D-Model is a beta feature for an Explicit LS-Dyna analysis." All papers / examples I found on SSI using LS Dyna are 3D models. Now I am wondering if a 2D analysis is a good choice in general ? Or should I rather opt for a full 3D simulation ? My main driver for a 2D analysis was simplifying the modelling and saving computing resources. But if I run into issues and can't find any documentation or resources, I have gained nothing. I want to note I am very new to Ls Dyna, and Ansys in general.

I am using Ansys workbench 2024R1 (not the student version). Grateful for any advice I can get! Thank you.

Joerg

[Ushnish Basu]

You can use 1 layer of solid elements and constrain them perpendicular to the layer to simulate plane strain

[Joerg]

That's a good suggestion. That should keep the modelling effort and computing resources comparable to a 2D analysis, but still utilizes full functionality of the 3D analysis.... May even give better results for the contact region at structure-soil interface....

Thank you!